sed '1d' not showing after piping echo $variable

vincix · 09-30-2016, 04:58 AM

I've got this script:

Code:

ps -ly | while
read c1 c2 c3 c4 c5 c6 c7 c8 c9 c10
do
echo $c6 $c7 $c8 | sed '1d'
done

I'm trying to delete the first line, the header, so that only the numbers are output. The thing is that sed there doesn't work. And I've no idea why. Normally if I do sed '1d' file.txt, then the first line won't show up. Why is it different here?

I know that echo normally turns columns into a single line with words separated by spaces, but here this is not the case. Echo will still show columns (that's another thing I don't really understand - maybe it's because of the while loop? Though it shouldn't be).

If I try echo $c6 | grep -E [[:digit:]], then I get the correct output. But I'd like to specifically delete the first line.

pan64 · 09-30-2016, 05:03 AM

sed definitely works, just your script was not constructed properly.

Code:

ps -ly | while read ...
do
echo | sed   <<< here echo will print one single line and sed will delete the first line
done

what you really need is either use sed before the while or after the loop:

Code:

ps -ly | sed 1d | while read
...

or

Code:

ps -ly | while read ...
do
...
done | sed 1d

(not tested)

vincix · 09-30-2016, 05:10 AM

So the reason why echo shows columns instead of lines is because it echoes a line at a time and the newline is added by the while loop, right?

pan64 · 09-30-2016, 05:18 AM

echo by default adds a newline at the end, it is not the while loop.
echo does not print columns but what you specified to print: $c6 $c7 $c8

c0wb0y · 09-30-2016, 06:41 AM

Try this:

Code:

[c0wb0y@centOS ~]$ ps -ly | tail -n +2
S  1000  2795  1964  0  80   0  2180 28871 wait   pts/1    00:00:00 bash
R  1000  3504  2795  0  80   0  1460 37227 -      pts/1    00:00:00 ps
D  1000  3505  2795  0  80   0   660 28871 sleep_ pts/1    00:00:00 bash

MadeInGermany · 09-30-2016, 08:32 AM

You can pipe into an explicit block with an extra read for the first line; the while will loop over the remainder

Code:

ps -ly | {
  read header
  while read c1 c2 c3 c4 c5 c6 c7 c8 rest
  do
    ...
  done
}

vincix · 09-30-2016, 08:49 AM

Quote:

Originally Posted by MadeInGermany

You can pipe into an explicit block with an extra read for the first line; the while will loop over the remainder

Code:

ps -ly | {
  read header
  while read c1 c2 c3 c4 c5 c6 c7 c8 rest
  do
    ...
  done
}

I was actually following this lynda course on bash, and I did have the solution, but I wanted to understand some basic stuff before looking at it and, of course, trying it myself.

The idea was to do the sum of all the numbers that showed under RSS and SZ. This is what they came up with:

Code:

#!/bin/bash
n=1

ps -ly | while
read c1 c2 c3 c4 c5 c6 c7 c8 c9 c10
do
        if ((n>1))
        then
                ((rss=rss+c8))
                ((sz=sz+c9))
                echo rss=$rss sz=$sz
        fi
((n++))
done

Which is still a little bit hard to thoroughly understand, even though I could easily reproduce it (they're few lines, so yeah

). What I don't understand, though, is, now that you've mentioned it, how the loop indeed skips the first line. The guy in the video says that he's actually using the "n" variable to do it. But I don't see how that affects the script.

rknichols · 09-30-2016, 08:57 AM

Quote:

Originally Posted by vincix

I've got this script:

Code:

ps -ly | while
read c1 c2 c3 c4 c5 c6 c7 c8 c9 c10
do
echo $c6 $c7 $c8 | sed '1d'
done

I'm trying to delete the first line, the header, so that only the numbers are output. The thing is that sed there doesn't work. And I've no idea why. Normally if I do sed '1d' file.txt, then the first line won't show up. Why is it different here?

The echo and sed commands are invoked separately for each iteration of the loop. Each echo command writes a single line, and each sed command deletes the first (and only) line that it sees.

If you want sed to filter the combined output from the loop, write it that way:

Code:

ps -ly | while read c1 c2 c3 c4 c5 c6 c7 c8 c9 c10
do
    echo $c6 $c7 $c8
done | sed '1d'

grail · 09-30-2016, 01:46 PM

Quote:

What I don't understand, though, is, now that you've mentioned it, how the loop indeed skips the first line. The guy in the video says that he's actually using the "n" variable to do it. But I don't see how that affects the script.

The magic is the 'if' statement. 'n' is originally set to 1 and when the first line is read, the one you want to get rid of, 'n' is still 1 so the 'if' skips this line as the test says 'n>1'.
After skipping the first lot of output, 'n' is increased before the next iteration of the loop, hence now 'n' is 2 or greater and so all the lines being processed now will enter the 'if' clause as they
pass the test.

Hope that helps.

c0wb0y · 09-30-2016, 07:46 PM

How about this:

Code:

ps -ly | awk 'BEGIN { RSS=0; SZ=0 }; NR>1 {print $2, $3, $4, $5, $6, $7, $8; RSS+=$8; SZ+=$9} END { print "RSS =", RSS, "SZ =", SZ }'